Note that although collocation of the remote client and the enterprise bean may allow the container to reduce the overhead of calls through the remote interface, because any optimizations based on collocation must be done transparently, such calls are still likely to be less efficient than calls made using the local interface.
Can some one please explain above?
Regards,<br />Sandesh<br />(SCJCP, SCWCD, SCBCD - 99%, OCP-1)<br /> <br />Either find a way or create one.
If your remote interface stub is for some reason being used in the same jvm heap as the enterprise bean itself (i.e. with the remote object representing the bean), the container still needs to have some intelligence in the stub which would identify the client as being local to the bean. Determining this would still be some level of overhead when compared to using the local interface where the container doesn't need to make any such checks. It directly calls the bean's local object.